Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
Salesforce and Nvidia unveiled Koa, a domain-specific reasoning model built for Salesforce's Agentforce ecosystem, at Salesforce's annual Dreamforce event, where Nvidia CEO Jensen Huang joined Salesforce CEO Marc Benioff…
A September 18, 2026 guide ranked 11 open-source agent harnesses for local LLMs by OSI-approved license, documented local runtimes, maintenance status, and safety controls, with OpenCode taking the top spot for documenti…
Anthropic CEO Dario Amodei and Nvidia CEO Jensen Huang clashed over AI safety regulation at Salesforce's Dreamforce event, where Amodei advocated for independent evaluators to get permanent access to frontier models whil…
GPT-6 Astra reverse engineered the custom challenge-code format of the country guessing game Guessterra purely from example tokens, without access to source code, according to a first-person account by the developer's br…
Anthropic said on Thursday that its Claude chatbot now leads 26% of the company's research and development work and collaborates with human staff on more than 90% of tasks, warning that models accelerating their own deve…
TypeSafe.ai released Probably 0.1, a small programming language for LLM workflows that is executed by a TypeScript interpreter and powered by Jev for judgments and probabilities. The language uses `feels` for confidence-…
MSI launched the PRO MAX EDGE AI+, a 4-liter desktop delivering up to 126 TOPS of AI performance for local AI workloads. The system runs on an AMD Ryzen AI Max+ 395 processor with sixteen Zen 5 cores, forty RDNA 3.5 comp…
A Hacker News user reported that their organization was allowlisted by Anthropic's Cyber Verification Program but is still being downgraded to Opus 4.8 when checking its own code for security problems. The user asked oth…
MiniMax-H3, an omni-modal generative model that combines multimodal context understanding with joint audio-visual generation in a shared latent framework, is the subject of a new evaluation examining whether it can reaso…
Oracle co-CEO Clay Magouyrk told employees at an internal town hall that the company had not found broad internal uses for generative AI as recently as a year ago, before an April-May rollout of OpenAI's ChatGPT Enterpri…
Scientists including David Bellamy of the Institute of Foundation Models and Ginkgo Bioworks CEO Jason Kelly say the risk of AI creating bioweapons is low, arguing that assembling a virus and sourcing raw materials, equi…
A developer built an AI code review tool that enforces conciseness in application code rather than through prompts, using a GitHub webhook, diff fetching, Claude, structured JSON findings, and a severity-based ranker tha…
Alibaba's Qwen team released Qwen3.8-Omni-Flash, which it calls its first omni-modal model built around agentic capabilities, accepting text, images, audio, and video and returning text. The model is available today as a…
Large language models write verbose, low-quality code comments because reinforcement learning rewards comments that preserve design rationale across agent sessions, not because the models lack capability, according to an…
The AI Village, a multi-agent experiment launched by the research initiative AI Digest, has grown from four agents running two hours a day to 27 agents running eight hours a day on shared goals including creating merch s…
An analysis of 14,767 arXiv evaluation-resource papers published between 2022 and 2026 found that LLM benchmarks are shifting toward action, interaction, and professional-use tasks, with LLM-based scoring increasingly us…
The new BioPhys-Bridge benchmark, a 500-case, 1,517-task dataset for physics-grounded biological reasoning, shows state-of-the-art LLMs top out at an evidence-retrieval F1 of just 0.360, with DeepSeek-V4-Flash leading an…
Alibaba's Qwen released Qwen 3.8 Omni Flash, a 3.8-billion-parameter model that natively integrates text, vision, and audio processing in a single low-latency architecture. The model is positioned to run fully local, rea…
Bend, a high-level programming language that compiles mathematically verified code to run natively and in parallel on both CPUs and GPUs without manual multi-threading, is pitched as a constraint layer for LLM agents gen…
A developer and commentator, appearing on the BBC, pushed back on claims from Sam Altman, Jensen Huang, and Bernie Sanders about AI risk, arguing that Altman's and Huang's reassurances that AI companies will self-police …